Professional Documents
Culture Documents
net/publication/303816864
CITATIONS READS
12 1,295
3 authors, including:
Bernhard Stauder
Nemak Linz GmbH
27 PUBLICATIONS 196 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Mechanical and thermomechanical fatigue properties of heat treated aluminium cast alloys View project
All content following this page was uploaded by Bernhard Stauder on 10 January 2020.
a r t i c l e i n f o a b s t r a c t
Article history: Improved understanding of foundry sand core properties is a key requirement for high precision cast-
Received 25 September 2015 ing process development. The present work demonstrates the potential to evaluate mechanical and
Received in revised form 1 June 2016 functional sand core properties using precisely acquired 3-point bending test load curve data applying
Accepted 4 June 2016
standard bending test geometries.
Four organic binder systems have been investigated. Further to bending strength and the elastic mod-
Keywords:
ulus, which can be directly derived from the load curves, a load curve pre-treatment to eliminate sample
Casting
settlement effects was applied for a corrected deflection and stiffness analysis. The consumed mechanical
Sand core testing
Load curve evaluation
work shows characteristic elastic and plastic work portions until fracture, which are specific for different
Work of fracture sand cores, respectively their condition. Dimensionless indicators to quantify core brittleness have been
Core deformation developed based on curve and on work parameters.
Quality control In general for mechanical evaluations of resin bonded sand cores, visco-plastic effects need to be con-
sidered, as for not fully hardened binder systems decreased strength, deflection and work of fracture
were observed at lower load speeds.
The benchmark results show that the load curve evaluation concept is a suitable tool to analyse foundry
sand core properties more sensitively.
© 2016 Elsevier B.V. All rights reserved.
Metal casting using sand cores allows to produce complex The development of new binder systems, such as chemically
components with internal geometries in high volumes, typical bond sand cores was intensively characterized by Bindernagel
examples are combustion engine cylinder heads and blocks. Fig. 1 (1983). A comprehensive collection of sand and binder systems
shows a cylinder head casting from Al and the required sand cores was given by Flemming and Tilch (1993). Mechanical properties
for the internal surfaces, while in Fig. 2 a preassembled sand core of sand cores are mainly qualitatively described and no property
package is shown prior to casting. related material laws are proposed.
The influence of technology, materials and process parameters In automotive foundries the most widely applied core manu-
on dimensional accuracy of near-net-shaped casting products has facturing processes are the organic Coldbox, Warmbox, and Hotbox
been systematically compiled by Campbell (2000). The foundry core production processes. Process descriptions are given by Langer
industry is forced to steadily improve process capabilities to facil- and Dunnavant (2011) and Brown (2010). Currently a substitution
itate light-weight, respectively thin walled castings, to contribute process of organic by inorganic binders having less odour develop-
to ever tightening emission legislation requirements (Lellig et al., ment is ongoing. However, because of different properties, process
2010). The importance of sand core technology development is relationships for such new binder types require increased atten-
highlighted. tion on the process control (Weissenbek et al., 2011). Recently
Czerwinski et al. (2015) reviewed the state of the art of foundry
core technology for several relevant organic and inorganic binder
systems used in various foundry applications.
∗ Corresponding author.
Summarized, fundamental requirements into sand core proper-
E-mail addresses: bernhard.stauder@nemak.com, bernhard.stauder@aon.at
ties are:
(B.J. Stauder).
http://dx.doi.org/10.1016/j.jmatprotec.2016.06.010
0924-0136/© 2016 Elsevier B.V. All rights reserved.
Page 65
B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196 189
Fig. 1. Illustration of a medium complexity car cylinder head in cast Al and the required sand cores to shape its cavities (Sobczyk, 2008).
Page 66
190 B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196
Fig. 3. Tensile and compressive forces on particles during 3-point bending (Wolff et al., 2013).
Fig. 4. Sand grain morphology of the used H32 silica sand and the according sieve analysis.
Table 1
Used binder systems for the experiments and main hardening parameters.
Coldbox (CB) Phenolic resin and isocyanate solution in 1:1 ratio. DMEA vapour gassing 15 s
Di-Methyl-Ethyl-Amine (DMEA)
Hotbox (HB) Phenol formaldehyde resin and aqueous ammonium nitrate solution in 5:1 ratio. Heated tool: 220 ◦ C/30 s
Warmbox type 1 (WB1) Furanic resin and aqueous phenol-sulfonic acid solution, its Al-salts and sulphuric acid Heated tool: 220 ◦ C/30 s
in 5.5:1 ratio.
Warmbox type 2 (WB2) Modified urea-furanic resin and aqueous phenol-sulfonic acid solution and its Al-salts Heated tool: 220 ◦ C/30 s
in 4:1 ratio.
for foundry sand cores. They demonstrated the improvement pointed out the limitation of using linear continuum mechanics for
potentials by applying modern testing technology with precise bonded granular materials when having plasticity. Improvements
load-displacement acquisition. A load velocity of 5 mm/min has were shown by Wolff et al. (2013), applying “Discrete Element
been determined to achieve comparable conditions to existing Method (DEM)” – simulations for a ceramic-polymer composite
bending test practice. material under different bending test conditions. For small loads
Ratke and Brück (2006) analysed load displacement curves they have shown that visco-elastic effects could be neglected for
from bending and compression tests to investigate the influence of the used acrylic binder. Moreover, initial settling against the load
granulometry and binder content on silica-Aerogel bond compos- points are important and must be considered. They illustrated the
ites. Thole and Beckermann (2009) researched the deformation of performance of DEM simulations to investigate bonded particle
phenolic-urethane no-bake elf-setting cores applying three-point compounds. A three-point bending DEM-setup is shown in Fig. 3.
bending test standards from metal testing. Elastic moduli close to Tarokh and Fakhimi (2014) compared measured and simulated
4 GPa in the as-produced samples and about 1 GPa after thermal bending test curves of silt stones. A satisfyingly good agreement
exposure were determined. was given identifying the initial deformation of the physical spec-
(Schacht, 2004; chap. Schacht&Hayes, Silica brick material imen. The early portion of the curve was influenced by closure of
properties) described the load-displacement curve evaluation for the space between the specimen and the loading platen and fur-
compression tests on refractory silica bricks. The relation between thermore a postulated closure of micro-cracks in the specimen.
the maximum slope tangent and the secant from zero to the max- Further to load curves, the introduced work into the sample can
imum load point was presented as a suitable criteria to describe be analysed also. Rice (1968) presented the “J-integral”-method
plasticity. (Schacht, 2004; chap. Bradt, Fracture of Refractories) to describe crack tip plasticity. An adoption of that principle will
Page 67
B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196 191
Fig. 5. Foundry sand core sample at bending test. The three supports are line sup-
ports with a top radius of 3 mm according to the German Foundrymen’s Standards.
Fig. 7. Bending strength results for all tested cores over storage time.
Fig. 6. Untreated high resolution stress – deflection curves. Core binder WB1 tested
after 4 h storage time. Fig. 8. Illustration of the load curve settlement correction pre-treatment method
based upon the maximum stiffness tangent. Additionally the secant to the maximum
strength is indicated.
be used in this work. The goal is to quantify the introduced work
from the testing machine and evaluate the elastic and plastic work
describes the handling strength of the sand cores, which is impor-
portions.
tant for their first manipulation after the core production. Testing
was performed 5 min after core production. The results after 1 and
2. Experimental methods 24 h represent the core properties at the beginning and end of a
typical core usage in foundry practice.
2.1. Used materials Mean values and standard deviations were determined for a
minimum of three samples.
The sand used for all samples was silica sand “H32” accord-
ing to the technical datasheet (Quarzwerke, 2009), being a widely 2.3. Three-point-bending-tests
used reference for binder testing purposes. Fig. 4 shows its grain
morphology and the grain size distribution. The bending tests were performed on a 5 kN universal test-
Experiments are based upon one sand type and four different ing machine (Zwick Roell, 2015) with a precise load-displacement
organic binder additions ranging from 1.2–1.9 wt.%. Table 1 defines measurement. A load velocity of 5 mm/min and a data acquisition
the used binder systems and the hardening parameters, which are rate of 10 Hz were set for the benchmarking.
similar to typical serial applications used for Al-castings. The lower support distance l is 150 mm and the cross section
side length a is 22.7 mm. The bending stress calculation from the
2.2. Sample production measured load F [N] according to continuum mechanics is shown
by Eq. (1).
The sand-binder mixtures were prepared in 5 kg batches using
3l
a laboratory stir mixer and then immediately used for core pro- =F× (1)
duction. A conventional 5 l core blowing machine (Roeper H5) 2a3
and a core box for three test bars per shot according to (VDG, A three-point bending test sample put on the machine supports
1974) were used. The effectively obtained test bar dimensions were according to the VDG-testing standard is shown in Fig. 5.
22,7 × 22,7 × 180 mm3 . The samples were stored and tested in in
standard room conditions with 30–50% relative humidity. 3. Results
Bending tests were conducted after storage times of 0, 1, 4 and
24 h. The storage times were chosen as they are representative The result section is structured to show the novel data acquisi-
for high volume engine casting production. The “0 h”-condition tion and the subsequent evaluation method leading to evaluation
Page 68
192 B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196
Fig. 9. Bending stress curves as function of beam deflection after 0 h, 1 h, 4 h and 24 h storage time. The binder systems are: (a): Coldbox—CB; (b): Hotbox—HB; (c):
Warmbox-1—WB1; (d): Warmbox-2—WB2.
of the sample brittleness. A final sequence will show the influence ities due to the granular nature of the sand cores can be observed.
of load velocity on the properties. Sample fracture occurs spontaneously after reaching the maximum
strength.
From Fig. 7 it can be seen, that HB- and WB-cores reach their
3.1. Data acquisition and bending strength evaluation
maximum bending strength at 1–4 h. After 24 h of storage, the
bending strength of WB1- and HB-cores is decreased by about 3%,
A typical example of acquired load curves is shown in Fig. 6
and for WB2-cores more significant, by 12%. The effect is generally
(WB1 binder system after 4 h storage time). The acquired bending
associated to humidity take-up after the samples are cooled down,
strength results are visualized for all tested conditions in Fig. 7.
leading to softened bonds.
The different curve onsets in Fig. 6 are caused by settlement
In contrast to that, CB cores show much lower bending strength
effects at the start of loading. However, the inclining curve shape
but a continuous increase over time due to ongoing polymerisation.
is similar for all tested samples, and highly linear. Some irregular-
Page 69
B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196 193
Table 2
Bending strength average values and standard deviation.
t CB HB WB1 WB2
b l3
E= = Cmax (2)
ε 4a4
Fig. 11. Elastic modulus results for all tested cores over storage time.
Fig. 10 shows beam deflection results at the maximum load level
(s ) and Fig. 11 the elastic modulus results. 3.3. Mechanical work evaluations
Beam deflection from pre-treated load-curves can be used as a
first indicator for core deformability. For hot hardening systems the The totally introduced work WM into the testing setup is dis-
deflection results are significantly below that of CB-cores, except played by the area below the untreated load-displacement curve.
for WB1-cores, developing the highest strength. In Fig. 12 an exemplary pre-treated load curve is shown. The char-
Interestingly, the elastic moduli of the different HB- and WB- acteristic points and the distinguishable elastic and plastic work
systems are very similar to each other. After 1 h storage time E areas I, II, III and IV are illustrated.
is about 2.7 GPa. The CB-cores exhibit about 30–50% lower elastic Work area I (Wpl i : initial plastic work) defines the area above the
moduli. After 24 h the CB-samples obtained a level of 1.7 GPa. All maximum stiffness tangent (P0 to PC ) and the load curve. This area is
systems exhibited 20% lower values at the 0 h condition compared dedicated to settlement effects of the sample towards the supports.
to the 1 h results. This, combined with initially higher deforma- This work portion is not contributing to the sample fracture.
bility underlines the importance of proper handling and storage Work area II (Ws pl : plastic work until max. stress) defines the
conditions for newly produced sand cores. area between the corrected load curve and the linear incline to P
Page 70
194 B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196
Fig. 13. Mechanical work of bending test samples. Initial and final plastic work and elastic work are separately indicated for each binder type over the storage time t.
Page 71
B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196 195
Table 3
Mathematical definition of the mechanical work regions I–IV.
Z
smax
X
smax
Region: I II III IV
Denomination: Wpli Ws pl Wel Wfpl
Z
sC
Zs Z
smax
0 sC s
FC · sC FC · sC Fmax2 Fmax 2
Added (+) or subtracted (−) elastic energy terms: − 2
+ 2
− 2Cmax
+ 2C max
–
WS
Table 4
Dependence of CB properties on the load velocity (low: 0.5 mm/min and high: 50 mm/min). Values are given as percentage of the 5 mm/min – reference results. Values for
0, 1 and 24 h storage time were evaluated for a minimum of 5 samples per data point.
Results in [%] of 0.5 mm/min – values bending strength deflection elastic modulus total work at mod. based br. work based br.
fracture index index
Load vel.
Storage time low high low high low high low high low high low high
Fig. 14. Modulus based brittleness index results for all tested cores over storage Fig. 15. Energy based brittleness index results for all tested cores over storage time.
time.
generally below the modulus based ones, more distinct for the
bending test setups only, because sample stiffness and elastic work CB-cores than for the hot hardened core systems. Investigating
are size dependent parameters. materials with some plasticity, the work based approach should
Generally, the more a load curve is bowed, the lower is the brit- be chosen, to include the final curve section also into the evalu-
tleness index. For a completely linear load curve it would reach ation. The modulus based evaluation reflects curve relations until
unity. Results for BM are shown in Fig. 14 and for BW in Fig. 15. the maximum strength level only.
The work based brittleness indices are slightly lower than the
modulus based ones. high brittleness indices of above 0.85 are given 3.5. Influence of load velocity
for HB- and WB-cores. The most brittle system from both evalu-
ations is the WB2-system. The bent curve shape and the higher Due to their high plasticity and time dependency of the results,
plastic work fraction lead to lower brittleness indices for the CB- CB cores have been chosen to study the influence of load velocity.
system. Especially at the 0 h-condition the values are below 0.5, Load rates of 0.5–5 and 50 mm/min have been applied. In Table 4
giving evidence for a high deformability of those cores. Practically the results are shown in relation to 0.5 mm/min load velocity, which
the need of a careful core manipulation to avoid deformations is was constant for the previously shown results.
obvious also from these evaluations. For the low load velocity of 0.5 mm/min, resulting in about
f
For CB-cores the final plastic work portion Wpl (Fig. 13) con- 1 min bending test time, strength and total work are significantly
tributes significantly to WS. The work based brittleness indices are decreased by 20–25% related to the nominal load velocity of
Page 72
196 B.J. Stauder et al. / Journal of Materials Processing Technology 237 (2016) 188–196
Page 73